Study On Effect Of Environmental Factors On Dynamic Characteristics Of Structure

The modal frequency obtained from long-term mass monitoring data is actually the frequency of the whole structure system. Therefore,in the field of bridge structures damage identification with modal frequency fingerprint, the way to eliminate the influence of environmental factors has become a hot topic so that we can get the real modal frequency of structures. In this paper, focusing on theidentification of modal frequencyby eliminatingthe effect of temperature, vehicle and other environmental factors, the ninth continuous beam of the New Yihe River Bridge is set as the research object, whose normalized modal frequency is obtained with nonlinear SVM (Support Vector Machine) regression model of modal frequency against environmental factors. The main work and achievements are as follows:(1) Effects induced by environmental factorssuch as temperature and vehicle load on modal frequency of structures is investigated:the effect of temperature change is caused by change of elasticity modulus and other material properties and change of statically indeterminate restriction of the whole structure, while effect of load such as vehicle, rain and snow, wind is due to change of vibration characteristics of the whole structures.(2)Application of support vector regression machine in structure health monitoring is discussed:The FEM model of New Yihe River bridge is built with ANSYS, the multi modal frequency is obtained by simulating change of vehicle load, which is characterized by the SVM regression model of modal frequency against four variables of vehicle load. The research reveals that the SVM regression model can predict the change of structural modal frequency precisely.(3) Based on the long-term monitoring data provided by vehicle health monitoring system and weighing system installed on New Yihe River bridge, nonlinear SVM regression model of double environmental temperature variables, four vehicle load variables and six environmental temperature and vehicle load variablesagainst modal frequency is built. The research reveals that smaller sampling interval and more characteristic variables can improve the regression accuracy and model mentioned in this paper can predict the frequency change with enough precision.(4) Apply normalization processing to the measured frequency data of New Yihe River bridge with regression model proposed in this paper. The research reveals that the normalized frequency is equippedwithapparentlynarrowerrange and better stability, so that better structural damage alarming system is achieved.